himself65/finance-skills
22 skills17.4k installs60.1k starsGitHub
Install
npx skills add https://github.com/himself65/finance-skillsSkills in this repo
1Yfinance Datayfinance-data is a Python-backed agent skill that standardizes how you fetch Yahoo Finance market data through the open-source yfinance library. Solo and indie builders use it when a ticker, earnings question, or “download stock history” request would otherwise spin up one-off scripts or brittle copy-paste from finance sites. The workflow starts by verifying yfinance in the environment, installing quietly if needed, then executing the right calls for quotes, fundamentals, options, dividends, and holder analytics. It is aimed at Claude Code, Cursor, Codex, and similar coding agents that can run shell and Python in a repo or sandbox. Use it during opportunity research, prototype dashboards, or backend jobs that need normalized financial series—not as investment advice or a licensed market-data feed. Triggers align with everyday builder language: stock price, financials, options chain, analyst targets, and compare stocks.1.3kinstalls2Stock CorrelationStock Correlation is a finance agent skill that routes user intent to four specialized analysis paths over historical equity prices. Solo builders researching trading ideas, fintech side projects, or portfolio hedges can ask in plain language—what moves with NVDA, how AMD correlates to NVDA, or a FAANG matrix—and get structured quantitative output instead of hand-rolling pandas scripts each time. Co-movement discovery scans curated sector and thematic peer universes; return correlation does deep pairwise stats; sector clustering builds full matrices with hierarchical groups; realized correlation stresses time-varying and regime-conditional relationships such as co-drawdowns. The skill expects Python 3.8+ and handles dependency bootstrap for yfinance, pandas, and numpy. It fits Grow-phase analytics for ongoing market monitoring, and Validate-phase pricing or scope conversations when correlation evidence informs niche or B2B fintech bets. It is analysis tooling, not investment advice; verify data vendor limitations and your own compliance boundaries before shipping user-facing features on top.1kinstalls3Options Payoffoptions-payoff is an agent skill for solo builders and indie traders who need a clear picture of options risk before acting. It turns leg definitions—strikes, premiums, expiry, and spot—into an interactive payoff widget with dashed expiry intrinsic curves and solid Black-Scholes marks at current DTE and IV. You can pick a template strategy or stack arbitrary legs in custom mode, then tune parameters with sliders while stats update for max profit, max loss, breakevens, and mark-to-model P&L at spot. The skill fits early trade design and fintech prototyping when you are explaining structures to a cofounder, validating a hedge, or pairing an agent with screenshots from common brokers. It does not place orders or stream live quotes; it visualizes structure and sensitivity so pricing decisions are explicit instead of guessed in chat.975installs4Generative Uigenerative-ui encodes Anthropic’s built-in generative UI guidelines so Claude can render `show_widget` HTML and SVG without calling `read_me` first. Solo builders shipping finance explainers, growth dashboards, or launch one-pagers inside claude.ai conversations get consistent charts, diagrams, and interactive widgets instead of broken styling or illegible dark-mode contrast. The skill stresses flat, compact visuals that blend with the host UI and mandates dual-theme tokens. Triggers cover visualize, chart, dashboard, flowchart, mockup, and explain-how-X-works with visuals. It is procedural design knowledge—not a code generator repo—optimized for inline widgets rather than full SPAs. Pair it when you need trustworthy visual answers alongside markdown prose.914installs5Hormuz StraitHormuz Strait is an agent skill for solo builders and indie traders who need fast, cited context on one of the world’s most important oil chokepoints without opening a dozen news tabs. When someone asks whether Hormuz is effectively open, how tanker traffic looks, or what a Gulf crisis might mean for Brent and freight insurance, the skill runs a single read-only HTTP call to the public Hormuz Strait Monitor dashboard API and turns the JSON payload into a structured briefing. It is aimed at Claude Code, Cursor, or Codex users shipping finance automations, research agents, or internal ops dashboards—not full brokerage platforms. Use it during early market research, before rebalancing energy exposure, or when validating whether a supply-chain story is still current. The workflow is deliberately minimal: curl, parse, summarize; no credentials, no writes, no trading execution.899installs6Telegram Readertelegram-reader is a read-only integration skill that uses tdl to pull messages from Telegram channels and groups you already follow. Solo builders doing market research use it to scan what niche finance and crypto channels are saying without building a custom Telegram bot or violating write-only compliance rules. Capabilities include listing chats with filters, exporting recent or time-bounded messages, and searching channels by name or type. Authentication is interactive once—QR or phone—then the session lives on disk for later agent runs. The skill is deliberately narrow: it does not post, join, or leave channels, which keeps blast radius small but means it cannot automate outreach. Prism tags it under Idea research with a Grow secondary path for ongoing news monitoring. Pair it with your own note-taking or alert workflow; it is an ingestion layer, not a trading or execution system.877installs7Discord Readerdiscord-reader lets solo traders and indie researchers pull context from Discord communities they already join, using opencli to attach to Discord Desktop over CDP instead of operating a bot account. The skill documents listing servers and channels, reading recent messages, searching keywords in the active channel, and glancing at online members—all strictly read-only for compliance and safety. Triggers cover market sentiment on tickers, crypto group monitoring, and channel discovery when financial news breaks in semi-private communities. Setup expects Discord running with remote debugging on port 9232 and opencli installed per the upstream project. It does not work on Claude.ai where network and binaries are restricted; target Claude Code or similar CLI agents. Builders should treat this as a research adjunct to public feeds, not authenticated exchange data. Pair with your own journaling or alerting workflow after you ingest what people are saying in server threads.857installs8Twitter ReaderTwitter Reader is a finance-oriented agent skill that pulls read-only data from X/Twitter for solo builders doing market and sentiment research. Through opencli it can load your home feed, search keywords with relevance or recency, inspect trending topics, browse bookmarks, fetch chronological user timelines, and read threads or long-form articles—all without write actions that could affect your account. Authentication piggybacks on an existing Chrome login to x.com via the Browser Bridge extension, which removes API key setup for quick fintwit checks. Use it when you want qualitative signal alongside price data: what analysts and retail are saying about a ticker, theme, or macro headline. It is deliberately not a trading execution or social posting tool; scope stays research and monitoring. Confirm you run a CLI-based agent with network and browser bridge support before relying on it in your workflow.857installs9Finance Sentimentfinance-sentiment is an agent skill that gives solo builders and indie traders fast, structured social and news sentiment for US-style tickers through the Adanos Finance API. It is read-only: your agent runs curl against the API with ADANOS_API_KEY set and returns normalized signals from Reddit, X.com, news, and Polymarket, including buzz, bullish percentage, mention or trade counts, and trend direction. Use it when you need quantitative answers to questions like how hot a symbol is on X this week, how Reddit mentions compare to news, or whether prediction-market activity on a name is spiking. It works on any agent platform that allows shell and outbound HTTP. It does not execute trades or modify accounts; it compresses hours of manual tab-hopping into one consistent snapshot for Idea-stage research or Grow-stage narrative checks.853installs10Earnings RecapEarnings Recap is a finance agent skill that generates post-earnings analysis for any ticker using Yahoo Finance via yfinance, aimed at solo builders and operators who need fast context after a report—not a full sell-side note. It assembles EPS actuals against expectations with surprise percentage, rolls up roughly four quarters of revenue, margins, and EPS, and quantifies how the stock moved on earnings day compared to its typical earnings-day volatility. That combination helps you answer whether a move was an overreaction, whether margins are improving, and how growth is tracking without opening five browser tabs. Triggers mirror natural chat queries such as ticker earnings recap, beat or miss, and what happened after GOOGL reported. Prerequisites are Python 3.8+ and yfinance, which the workflow can install if absent; no API keys are required. Reference material documents yfinance methods for earnings history and financial statements. Use it in Claude Code, Claude.ai, or other agents when you want a repeatable earnings summary script pattern rather than one-off web scraping.788installs11Earnings Previewearnings-preview is a finance-skills agent skill that builds a pre-earnings briefing for a stock symbol using Yahoo Finance through the yfinance Python library. Solo builders, indie traders, and founders writing market commentary can see the next earnings window, street consensus on EPS and revenue, recent beat-or-miss history across four quarters, analyst sentiment buckets, and price targets without signing into a brokerage API. Triggers are natural language ticker questions and phrases like earnings preview, consensus estimates, and earnings season preview. Python 3.8+ is required; yfinance installs on demand. The skill ships a references/api_reference.md for calendar and estimate methods, and runs on Claude Code, Claude.ai, and other agents that can execute the bundled workflow.770installs12Estimate Analysisestimate-analysis is a finance agent skill that deep-dives analyst consensus for any ticker using Yahoo Finance through the yfinance Python library. It surfaces EPS and revenue estimate distributions across near-term quarters and fiscal years, then tracks how those numbers moved over 7, 30, 60, and 90-day windows with upward versus downward revision counts. Solo builders and indie investors use it to sanity-check bullish or bearish narratives, compare growth expectations to industry, sector, and S&P 500 baselines, and review historical accuracy patterns before acting on a thesis. Triggers include natural phrases like estimate analysis for AAPL, EPS revisions, consensus changes, and estimate momentum. Setup is lightweight: Python 3.8+, optional auto-install of yfinance, and no authentication. The skill ships API reference material for estimate-related ticker methods so agents call the right DataFrame fields instead of improvising scraping.757installs13Stock Liquiditystock-liquidity is a finance agent skill that walks solo builders and indie quants through liquidity due diligence on individual equities using Yahoo Finance data through yfinance. Quoted mid prices rarely match executable prices; spreads, thin average daily volume, and market impact can dominate returns especially on small caps or penny names. The skill is invoked when users ask how liquid a ticker is, what bid-ask looks like, whether a large block can trade without moving the market, or how turnover and illiquidity metrics compare across names. It organizes analysis across spreads, volume profiles, order book depth proxies, impact estimates, and turnover ratios so agents produce consistent research memos rather than hand-wavy liquidity opinions. The README stresses research and education only—not investment advice—and notes yfinance is third-party. Fit for CLI or agent tooling that generates pre-trade checklists before validate-scope position sizing or operate-era execution monitoring.754installs14Funda DataFunda Data is an agent skill that routes financial questions through Funda AI’s MCP server for synthesized analyst-grade answers and through the REST API when you need structured raw feeds. Solo builders use it during idea and validate work to compare sectors, read earnings and filing context, and pull line-item or market data into prototypes, dashboards, or trading-adjacent agents. The skill explicitly prefers MCP for narrative research—beat-miss decomposition, estimate revisions, competitive positioning—and falls back to REST for intraday prices, options greeks, ownership, news calendars, and bulk downloads. It fits indie teams shipping internal research copilots or validation spreadsheets without wiring every endpoint by hand. You still need appropriate Funda credentials and should treat outputs as research assistance, not investment advice.738installs15Startup Analysisstartup-analysis is an agent skill that produces a multi-perspective report on any startup by simulating how a VC investor, a job applicant, and a CEO/founder would read the same public signals. Solo builders use it when choosing co-founders, evaluating competitors, deciding whether to join an early team, or sanity-checking a market before they validate or build. The skill explicitly relies on web search to gather fresh public context, then structures findings so strengths in one lens (e.g., investable growth) do not hide risks in another (e.g., weak employment equity or culture). It fits Prism’s journey before heavy build spend: Idea research and Validate scope, and it can inform Grow and Operate decisions when reassessing partners or acquirers. It is planning and judgment support—not legal due diligence or proprietary data-room analysis.679installs16Etf PremiumETF Premium is a finance agent skill for CLI-capable coding agents that quantifies how far an ETF’s market price sits above or below its net asset value, using Yahoo Finance as the data source. It answers trader-style questions—whether SPY trades rich, which bond ETFs are discounted, or why an ETF jumped more than its holdings—and adds comparison and screener modes plus a gamma-squeeze decomposition path that separates NAV-driven moves from structural premia and discusses dealer gamma and convergence timing. The readme positions it for Claude Code and Codex-style environments with Python, auto-installing yfinance when needed. For Prism’s indie audience, it suits personal investment research, content ideas, or validating fintech side-project features without building a Bloomberg terminal. It is not a journey-wide methodology skill; invoke it when ETF mispricing questions appear in chat.675installs17Sepa StrategySEPA Strategy is an agent skill that walks solo builders and indie traders through Mark Minervini’s Specific Entry Point Analysis—a disciplined framework for growth and momentum stocks. Invoke it when conversation touches SEPA, superperformance, Stage 2 uptrends, VCP, pivot breakouts, or questions like whether a setup is buyable. The skill sequences stage analysis (Stage 2 only), an eight-condition trend template, fundamental grading, consolidation pattern ID, entry geometry, position sizing, and bull/choppy/bear market adjustments. It is procedural knowledge packaged for Claude Code, Claude.ai, and other agents, with no special setup beyond bringing a ticker and context. Use it to replace gut-feel entries with explicit checklists and risk math, whether you trade personally or prototype screening assistants. It does not fetch live quotes by itself; pair it with your data source or broker workflow when you need numbers.661installs18Saas Valuation Compressionsaas-valuation-compression is a finance research skill for solo and indie builders (and small teams) who need defensible narratives around SaaS funding and multiples—not gut feel from a single headline round. It pulls funding history through web search, computes ARR-based valuation multiples at each round, and compares compression or expansion using a repeatable framework spanning data gathering, metric math, cause attribution, charts, and a tight written summary. Invoke it when you are benchmarking competitors, prepping investor questions, or explaining why a company’s multiple moved between rounds. It runs on common agent platforms and expects network search plus chart rendering; install via the finance-market-analysis plugin or the single-skill package when you want this workflow without the full bundle.650installs19Yc Readeryc-reader is an agent skill that wires your coding agent to the public yc-oss/api dataset so solo builders can research the Y Combinator portfolio without spreadsheets or manual site scraping. It fits the idea phase when you are exploring fintech, AI, B2B, or other spaces and need batch lists (e.g., W25), industry breakdowns, tag filters, hiring signals, and lightweight search over company names and descriptions. The API is read-only static JSON—no authentication—so the skill focuses on correct endpoints and jq-style filtering rather than write operations. It is especially useful for CLI agents that can run curl on the network; it is not aimed at sandboxed chat UIs that block outbound calls. Pair it with your own notes on differentiation and validate-phase scoping; it accelerates factual landscape scans before you prototype or pitch.650installs20Linkedin ReaderLinkedIn Reader is a read-only finance research skill that pulls LinkedIn home feed posts and job listings through opencli and a Chrome Browser Bridge extension, reusing your logged-in session instead of API keys. Solo builders and finance-minded operators use it to scan analyst commentary, professional sentiment, and quant or finance hiring trends while staying inside agent workflows. Write actions—posting, likes, connects, messaging—are intentionally out of scope to reduce account risk and compliance noise. It fits CLI-based agents such as Claude Code, not Claude.ai’s restricted network sandbox. Treat it as a lens on professional-network signal during research and occasional growth or hiring checks, not a marketing automation tool.640installs21Skill CreatorSkill Creator is a meta agent skill for solo and indie builders who treat procedural knowledge as product: it walks you from intent (“create a skill”, “turn this into a skill”, “benchmark this skill”) through structured SKILL.md authoring, reference file layout, quality rubrics, eval runs, and iteration until behavior is stable in Claude Code, Cursor, Codex, or similar clients. It emphasizes exhaustive when-to-use triggers, explicit defaults, step exit gates, and pulling long reference material out of the main file so invocation stays fast and deterministic. Use it when you keep repeating the same multi-step ritual—finance workflows, deploy checklists, review gates—and want the agent to load the same playbook every time. It is not a substitute for domain skills themselves; it is the factory that produces them. Expect to invest in evals when quality matters, and to align triggers with real user phrasing so discovery and AEO-friendly summaries match how you actually ask for help.573installs22Company ValuationCompany Valuation is an agent skill for solo builders and indie operators who want a fundamentals-driven view of a public company without opening a spreadsheet first. It automates a triangulated valuation workflow: historical and estimate data from yfinance, a five-year discounted cash flow model, relative multiples against a small peer set, and optional segment-level SOTP when the filer reports distinct businesses. Output is oriented toward decision-making—blended implied share price, method weights, sensitivity tables, and scenario bands—so you can sanity-check narratives like “is this overvalued” before pricing a side project’s TAM story or sizing a personal position. It runs in CLI agents with Python 3.8+, auto-installing yfinance, numpy, and pandas on first use, with optional fallback to a funda-data provider skill. Best when you already have a ticker and need structured assumptions you can challenge, not when you need private-company or pre-IPO modeling without public data.561installs